The author discusses the implications of a heterogeneous logit model f
or brand positioning. The methodology presented is a restricted versio
n of a mixture-of-logits model and obtains brand positions on a produc
t-market map and the distribution of preferences across households whi
le accounting for the effects of marketing variables on household bran
d choice behavior. The restriction involves imposing a factor structur
e on the covariance matrix of the distribution of intrinsic brand pref
erences. An empirical application of the methodology is presented usin
g A.C. Nielsen household-level scanner panel data on the purchases of
liquid laundry detergents. The results indicate that the proposed mode
l provides a better fit to the data than the unrestricted mixture-of-l
ogits model or the Choice Map methodology.